www.impactjournals.com/oncotarget/ Oncotarget, Vol. 6, No. 13

Complementary genetic screens identify the E3 ubiquitin ligase CBLC, as a modifier of PARP inhibitor sensitivity

Jessica Frankum1,*, Pavel Moudry2,*, Rachel Brough1, Zdenek Hodny3, Alan Ashworth1, Jiri Bartek2,3 and Christopher J. Lord1 1 The CRUK Gene Function Laboratory and Breakthrough Breast Cancer Research Centre, The Institute of Cancer Research, London, UK 2 Danish Cancer Society Research Center, Strandboulevarden, Copenhagen, Denmark 3 Institute of Molecular Genetics, Academy of Sciences of the Czech Republic, Videnska, Czech Republic * These authors contributed equally to this work Correspondence to: Christopher J. Lord, email: [email protected] Correspondence to: Jiri Bartek, email: [email protected] Correspondence to: Alan Ashworth, email: [email protected] Keywords: DNA damage response, ubiquitin-proteasome system, RNA interference screens, PARP inhibitors, CBLC Received: February 03, 2015 Accepted: February 20, 2015 Published: March 18, 2015

This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Abstract Based on a series of basic, preclinical and clinical studies, the Poly (ADP- ribose) Polymerase 1 (PARP1) inhibitor, olaparib, has recently been approved for use in ovarian cancer patients with BRCA1 or BRCA2 mutations. By identifying novel predictive biomarkers of tumour cell sensitivity to olaparib, it is possible that the utility of PARP inhibitors could be extended beyond this patient subgroup. Many of the known genetic determinants of PARP inhibitor response have key roles in DNA damage response (DDR) pathways. Although protein ubiquitylation is known to play an important role in regulating the DDR, the exact mechanisms by which this occurs are not fully understood. Using two parallel RNA interference-based screening approaches, we identified the E3 ubiquitin ligase, CBLC, as a candidate biomarker of response to olaparib. We validated this observation by demonstrating that silencing of CBLC causes increased sensitivity to olaparib in breast cancer cell line models and that defective homologous recombination (HR) DNA repair is the likely cause. This data provides an example of how defects in the ubiquitin machinery have the potential to influence the response of tumour cells to PARP inhibitors.

Introduction defects [3, 4], an observation that has also been confirmed in clinical trials [2, 5]. It seems likely that the sensitivity of The poly (ADP-ribose) polymerase, PARP1 plays a BRCA defective tumour cells to PARP inhibitors is caused key role in the repair of damaged DNA [1]. Upon binding by their characteristic defect in repair of DNA double damaged DNA, PARP1 uses β-NAD+ as a co-factor to strand breaks (DSBs) by homologous recombination (HR), synthesise poly (ADP-ribose) chains on a series of target a process controlled by BRCA1, BRCA2 and the DNA proteins [1]. This PARylation of PARP1 substrates initiates recombinase RAD51 [3]. Although various mechanisms the localisation of a series of DNA repair mediators to the to explain this synthetic lethality have been proposed, one site of DNA damage, before autoPARylation of PARP1 hypothesis is that PARP inhibitors restrict the release of causes its release from DNA [1]. Over the past few years, PARP1 from damaged DNA [6]. The DNA lesion that a number of small molecule inhibitors of PARP1 have results from the trapping of PARP1 on DNA likely stalls been developed for the treatment of cancer [2]. Pre-clinical DNA replication forks and requires functional HR for its work demonstrated that these inhibitors cause synthetic repair [6, 7]. lethality in tumour cells with BRCA1 or BRCA2 gene In addition to defects in BRCA1 and BRCA2, www.impactjournals.com/oncotarget 10746 Oncotarget preclinical work has suggested that alterations in a series describe these parallel approaches, as well as functional of additional genes also modulate the response to PARP studies that together pinpoint the ubiquitin ligase CBLC inhibitors. In totality these efforts have identified a number as a previously unrecognized factor whose depletion is of candidate predictive biomarkers of tumour cell response synthetically lethal with PARP inhibition. including ATM, ATR, CHEK1, CHEK2, DSS1, RAD51, NBS1, IPMK, NAMPT, ERCC1, the Fanconi anaemia Results complementation genes, PTEN and the TMPRSS2– ERG and EWSR1-FLI1 translocations (reviewed in [2]). Recently, we also described a genome-wide RNA Genome-wide PARPi shRNA screen identifies interference screen that identified a compendium of novel candidate olaparib sensitisation genes involved in PARP inhibitor sensitivity-causing genes, including the the control of ubiquitylation kinase-coding gene CDK12 [8]. The majority of PARP inhibitor sensitivity genes identified to date have a known role in DNA double To investigate the possibility that proteins involved strand break repair (DSBR). In part, the DSB signalling in the ubiquitylation machinery modulated the tumour and repair process is controlled by ubiquitylation, the cell response to a clinical PARP inhibitor, olaparib [3, post-translational, covalent attachment of 76 amino- 12], we first reanalyzed data from a previously published acid ubiquitin chains to target proteins (reviewed in genome-wide olaparib sensitization genetic screen [8]. [9,10]). Ubiquitylation is mediated via the coordinated In this screen, an olaparib resistant, BRCA1, BRCA2 activity of E1 (ubiquitin activating), E2 (ubiquitin- and p53 wild-type breast tumour cell line, MCF7, was conjugating) and E3 (ubiquitin ligase) enzymes and can transduced with a lentiviral library encompassing 57,540 be reversed by the activity of de-ubiquitylating enzymes short hairpin (sh)RNA expression constructs designed (DUBs) (reviewed in [9]). The role of these enzymes in to target 16,487 unique human protein-coding genes. DSB detection and repair is best exemplified by their The virally transduced cell population was subsequently involvement in the recruitment of key DSBR proteins divided into two cohorts, one exposed to olaparib for to the site of DNA damage. For example, DNA double two weeks, the other exposed to the drug vehicle. By strand breaks are recognised by the MRN (Mre11–Rad50– comparing shRNA frequencies in olaparib vs. vehicle NBS1) complex, an event that leads to the activation of surviving cell populations after two weeks, we identified signalling kinases that phosphorylate the histone H2AX 2,208 different genes whose gene silencing was predicted on chromatin that flanks the site of DNA damage. This in to result in enhanced olaparib sensitivity (those shRNAs turn enables the recruitment and phosphorylation of the that gave an olaparib sensitization Z score <-2) [8]. By DSBR mediator, MDC1. Once located to a DSB, MDC1 is re-annotating the candidate olaparib sensitisation genes itself phosphorylated; the phosphorylated amino acids on identified in this screen using pathway annotation tools MDC1 are bound by the E3 ligase RNF8, which initiates such as KEGG, we identified a series of genes implicated a series of ubiquitylation events that ultimately recruit in the control of ubiquitylation that were also implicated the E3 ligase RNF168. Together the RNF8 and RNF168 in olaparib sensitivity (Table 1). This gene list included ubiquitylation events enable the recruitment and retention both E1 Ubl-activating enzymes, E2 Ubl-conjugating of a series of DSBR factors including BRCA1, which enzymes as well as E3 Ubl-protein ligases and included also has E3 ligase activity and drives DSBR by HR, and known determinants of PARP inhibitor sensitivity such as 53BP1, whose activity drives DSBR via an alternative BRCA1 [3] as well as the UBA1 E1 enzyme coding gene, process known as Non Homologous End Joining previously reported to be required for responses to IR and (reviewed in [9, 10]). In addition to the role of E1, E2 replication stress in human cells (Supplementary Figure 1) and E3 enzymes in these processes, a recent systematic [13]. Amongst these olaparib candidate sensitivity genes, analysis of DUBs, which remove ubiquitin residues from we also noted RNF168, which is known to play a key role proteins, has highlighted the role of these enzymes in in the recruitment of a series of DSBR factors [9, 10, 14] DSBR and the maintenance of genomic integrity [11]. as well as a series of DUB enzyme-coding genes. These Given the known role of ubiquitin metabolism latter genes included BAP1, required for histone H2A in DSBR and the response to PARP inhibitors being deubiquitylation [15], BRCC36 (BRCC3) which has been determined by this process, we assessed the possibility associated with the in vitro response to the clinical PARP that additional genes involved in ubiquitin metabolism inhibitor rucaparib [16] and USP7S, whose gene product is might alter the tumour cell response to PARP inhibitors. involved in controlling the p53 response to DNA damage To do this, we have performed a high-content microscopy- [17]. based genetic screen for ubiquitylation-related genes, along with parallel analyses of data from a recent genome- wide RNA interference screen and integrated these two complementary datasets in this study. Below, we www.impactjournals.com/oncotarget 10747 Oncotarget Table 1: Genes implicated in Ubl metabolism identified as candidate olaparib sensitivity genes in [8]. The extent of olaparib sensitisation is indicated by the median Z score. Negative Z scores indicate a sensitization effect.

www.impactjournals.com/oncotarget 10748 Oncotarget www.impactjournals.com/oncotarget 10749 Oncotarget Table 2: Genes implicated in olaparib sensitivity as scored by automated micronuleation analysis. Hits are defined as those that showed more than 31 MN in the presence of Olaparib and 15 or less MN in DMSO. MICRONUCLEATION HITS 4 of 4 (3) 3 of 4 (14) 2 of 4 (104) CUL7 ANAPC1 ANAPC2 HGS RNF149 MAP3K7IP2 ASB6 ANAPC7 HIC1 RNF182 MYO1E ATR ANKRD13A HLTF RNF187 DHX9 ASB11 JARID1A RNF19A ERLIN1 ASB14 KALRN RWDD2A MDM1 ASB15 KBTBD6 SENP2 POLK ASB17 KBTBD7 SENP5 RNF17 BAG1 KIAA0999 SENP8 RNF4 BCL6B KIAA1333 SHPRH STAMBPL1 BIRC2 LNX2 SMURF2 UBE2H BMI1 LRPPRC SQSTM1 UBQLN1 BTBD14A MARCH10 STUB1 ZBTB3 CAND2 MARK2 SUGT1 ZNF395 CBLL1 MKRN2 TBK1 CCNF MLL2 TEX13A CDH1 NOSIP TMEM183A COPS3 NSD1 TOM1 COPS5 NSFL1C TOPBP1 CSMD3 NSMCE2 TRAF3 DDB2 OTUD5 TRAF5 DERL2 OTUD7A TREX2 DIP2C PARD6B TRIM23 DSG1 PHF12 TRIM8 DTX4 PHF21A TTC3 EIF2AK4 PHF7 UBL3 EPN3 POLI UBL4B FAM100A PSMA5 UBXD5 FAU PSMA8 USP45 FBXO34 PSMF1 VSP13A FBXW11 PYGO2 WDR24 FZR1 RB1CC1 WDR32 G3BP1 RCBTB1 XAF1 GZF1 RFPL2 ZBTB17 HADHA RFWD2 ZNFX1 HECTD1 RNF139 www.impactjournals.com/oncotarget 10750 Oncotarget A parallel high-content microscopy-based a determinant of olaparib sensitvity (DE Z score of -2.15, genetic screen identifies ubiquitylation-related Table 1). determinants of PARP inhibitor sensitivity Using manual scoring, no single gene scored in all four transfections but we identified two genes where siRNA caused aberrations in the examined nuclear In parallel to the genome-wide genetic screen, we parameters in the olaparib exposed cells in three out of also carried out a focused high-content genetic screen four transfections; the known olaparib sensitivity gene to identify genes associated with ubiquitin metabolism ATR (ATR serine/threonine kinase) and FBXO5 (F-box that are synthetically lethal with PARP inhibition. In this protein 5, Table 3) a new olaparib-sensitizing factor screen, the human U2OS osteosarcoma-derived cell line identified in our present screen. siRNA targeting 21 genes was plated on an siRNA array designed to target 1346 caused nuclear defects in the olaparib exposed cells in human genes involved in the ubiquitin-metabolism, two out of four transfections (Table 3). Of these genes, the proteasome system and genes encoding zinc-finger we noted that the siRNA designed to target the E3 ligase proteins [13]. Following siRNA transfection, cells were coding gene CBLC (Cbl proto-oncogene C [18]) caused exposed to either olaparib or the drug vehicle, DMSO for pronounced nuclear aberrations in olaparib-exposed cells three consecutive days (Figure 1A). At this point, cells but not in control DMSO-exposed cells (Figure 1B). were fixed, nuclei stained with Hoechst and images of shRNA designed to target CBLC also caused one of the cell nuclei were automatically acquired using wide-field most profound olaparib sensitisation effects in the MCF7 fluorescence microscope. We estimated the effect of each genome-wide shRNA interference screen (Z score -7.67, siRNA on olaparib sensitivity and genomic instability Table 1). By comparison, shRNA targeting BRCA1 gave using two approaches. First, we counted micronuclei an olaparib drug sensitisation Z score of -4.38. formation as an estimate of genomic instability caused by PARP inhibitor exposure using an automated ImageJ- CBLC silencing causes a HR defect based programme, and compared micronuclei frequency in olaparib and vehicle exposed cells (Table 2). In parallel, we also used manual scoring of nuclear and mitotic Given the profound effect of shRNA targeting CBLC defects (multinucleation, irregularity of nuclear shape and in the genome-wide genetic screen in MCF7 cells and anaphase chromosome bridges) (Table 3). the effect of siRNA targeting CBLC on olaparib-induced In both sets of analyses, we used our previously nuclear defects in U20S cells, we assessed the extent of established high-content microscopy-based screening PARP inhibitor sensitivity caused by CBLC silencing in strategy [13,14] and assessed the phenotypes in four dose-response clonogenic survival experiments. When independent transfections, imaging approximately 150 compared to a control, non-targeting shRNA expression cells that usually occupy each individual siRNA spot. construct, two different CBLC shRNA expression Across the entire screen, our automated analysis examined constructs caused a profound and significant increase in the average number of micronuclei for each of the olaparib sensitivity in MCF7 cells (Figure 2A, olaparib computer-identified multicellular siRNA spots. In control survival in shCONTROL targeted cells vs. shCBLC DMSO-exposed cells this overall micronuclei score was targeted cells, p<0.0001 ANOVA, and Figure 2B). We 2.63, while in control olaparib-exposed cells, this overall found the extent of olaparib sensitivity caused by CBLC average score was 15.35 (baseline average score). We shRNA to be equivalent to that caused by an shRNA considered as “hits” those siRNAs that elicited a greater expression construct designed to target BRCA1 (Figure than two-fold increase (30.7 or more micronuclei/spot) in 2A). We also found that siRNA reagents designed to target the frequency of micronuclei when exposed to olaparib, CBLC caused olaparib sensitisation in MCF7 cells as well but did not cause an alteration in micronuclei frequency as in a second breast tumour cell line model, HS578T in DMSO exposed cells. Using this approach, the (Figure 2C,D,E), suggesting that these effects were neither automated imaging analysis identified three genes (CUL7, restricted to the method of RNA interference used nor to MAP3K7IP2, MYO1E) that enhanced olaparib-induced the MCF7 cell line model used in the original genome micronucleation in all four transfections, 14 genes with wide shRNA screen. three out of four transfections scoring positively, and One key molecular determinant of tumour cell 104 genes where two out of four transfections elicited response to PARP inhibition is the ability to localise the micronucleation in response to PARP inhibitor exposure DNA recombinase RAD51 to the site of DNA damage, (Table 2). An example of the PARP inhibitor induced a critical event in HR repair that can be monitored by micronucleation phenotype is shown in Figure 1B, where the immunodetection of DNA damage induced RAD51 the effect of the siRNA targeting ANAPC1 (anaphase foci [3, 19]. We assessed the RAD51 response in human promoting complex subunit 1), an E3 ubiquitin ligase tumour cells transfected with CBLC siRNA, as well coding gene is shown. We noted that ANAPC1 also scored as the formation of γH2AX foci, a marker of histone in the genome-wide shRNA olaparib sensitivity screen as H2AX phosphorylation that is associated with DNA DSB formation [20]. Whilst control-transfected MCF7 cells www.impactjournals.com/oncotarget 10751 Oncotarget Table 3: Genes implicated in olaparib sensitivity by manual scoring of nuclear and mitotic defects. MANUAL SCORING HITS 4 of 4 (0) 3 of 4 (2) 2 of 4 (21) ATR ANUBL1 FBOX5 ARIH1 AURKA CBLC CCNF CDC20 CDC27 CUL1 DDB1 DDB2 DHX9 DPF2 FBXW7 PEX10 PHF21A PIAS4 RNF17 RNF4 SKP1 SPSB2 UBQLN3

Figure 1: Focused siRNA screen for Ubl modifiers of the PARP inhibitor response. A. Screening procedure. U2OS cells were plated on siRNA arrays and 24 h later treated with 10 uM Olaparib or vehicle (DMSO). After 3 days of cultivation cells were fixed, cells nuclei were stained with Hoechst and images corresponding to each siRNAs acquired using automated routine. Scoring was done in two ways; first using automated routine in program ImageJ number of micronuclei was determined. Manual scoring of nuclear and mitotic defects was employed as another way to determine which siRNA knockdowns sensitize cells to Olaparib. B. Images of Hoechst stained cell nuclei from siRNA screen illustrating the effect of siRNA targeting CBLC. www.impactjournals.com/oncotarget 10752 Oncotarget Figure 2: CBLC gene silencing causes PARP inhibitor sensitivity. A. Clonogenic dose-response survival curves from MCF7 human breast cancer cells infected with shRNA expression constructs targeting either BRCA1 or CBLC. Transduced cells were subsequently exposed to olaparib for 14 days at which point cell colonies were counted. Two different shRNA expression constructs targeting CBLC, shCBLC_1 and shCBLC_2, were used.****ANOVA p value <0.0001 for the dose response curves in either shCBLC_1 or shCBLC_2 transduced cells vs. shCONTROL transduced cells. Data from shBRCA1 transduced cells is shown as the positive control. B. Bar chart illustrating CBLC RT-PCR data from MCF7 cells expressing CBLC cDNA and transduced with shRNA expression constructs. *Student’s t test p value <0.05 for CBLC expression compared to shCONTROL tranduced cells. C. and D. Olaparib dose-response survival curves from MCF7 (C) or HS578T (D) human breast cancer cells transfected with siRNA targeting either BRCA2 or CBLC. Cells were transfected with siRNA and 48 hours later exposed to olaparib for a subsequent six days.****ANOVA p value <0.0001 for the dose response curves in siCBLC transfected cells vs. siCONTROL transfected cells. Data from siBRCA2 transfected cells is shown as the positive control. E. Bar chart illustrating CBLC RT-PCR data from MCF7 cells expressing CBLC cDNA and transfected with siRNA. * Student’s t test p value <0.05 for CBLC expression compared to siCONTROL transfected cells. Where error bars are shown these represent the standard error of the mean (SEM) from three independent experiments. www.impactjournals.com/oncotarget 10753 Oncotarget exhibited a clear increase in both RAD51 and γH2AX was moderately increased in cells transfected with siRNA foci in response to ionising radiation (IR) (Figure 3A and targeting either CBLC or BRCA2 (Figure 3B), compared B), cells transfected with either CBLC or BRCA2 siRNA to those transfected with a control siRNA. exhibited a clear reduction in the frequency of cells Taken together, these results suggested that a CBLC exhibiting a RAD51 foci response (Figure 3A and B). In defect could cause a reduction in the ability to repair DNA addition to the RAD51 defect, we also noted that 24 hours by homologous recombination (HR). To directly assess after the initial exposure to IR, the frequency of cells still the effect of CBLC silencing on HR, we estimated HR exhibiting γH2AX foci, and presumably unrepaired DSBs, activity using a synthetic DNA HR substrate, DR-GFP,

Figure 3: CBLC gene silencing causes a homologous recombination defect. A and B. Bar charts indicating the frequency of cells with greater than five RAD51 (A) and γH2AX (B) nuclear foci in response to exposure to 10 Gy ionising radiation and CBLC siRNA. CBLC siRNA reduced the frequency of cells with RAD51 foci * p<0.05 Student’s t test vs. control siRNA transfected cells at the same time point. Data from siBRCA2 transfected cells is shown as a positive control. C. Bar chart indicating GFP signal generated by repair of the DR-GFP DNA substrate. Data shown is normalised to the % of GFP positive cells in a mock-transfected sample. CBLC siRNA reduced the frequency of GFP+ve cells *p<0.05 Student’s t test vs. control siRNA transfected cells. Data from siBRCA2 transfected cells is shown as a positive control. Where error bars are shown these represent the standard error of the mean (SEM) from three independent experiments. www.impactjournals.com/oncotarget 10754 Oncotarget which generates a GFP signal once an experimentally of EGFR results in CBLC being recruited to the receptor induced DSB has been repaired by HR [21]. We found that where CBLC is phosphorylated. The interaction between the repair of the DR-GFP substrate by HR was impaired CBLC and EGFR leads to the attenuation of MAP kinase in both BRCA2 and CBLC siRNA transfected cells signalling [22]. Likewise, CBLC enhances ubiquitylation compared to control transfected cells (Figure 3C, p<0.05 and degradation of the oncoprotein RET [26, 27]. CBLC by student’s t-test), supporting the hypothesis that loss has also been shown to form a complex with HIC5 of CBLC activity causes a HR defect, a phenotype that (Hydrogen peroxide-inducible clone 5) and the heat shock could explain the PARP inhibitor sensitivity also caused protein HSP27. These latter interactions are thought to by loss of CBLC function. We also assessed the effect be required for the ubiquitylation of NOX4, a protein on progression of the MCF7 breast cancer cells through involved in the control of reactive oxygen species [28] the cell cycle in response to CBLC silencing. In both and diverse stress-induced cellular responses including the presence and absence of IR exposure, we found that oncogene-induced cell senescence [29]. It remains to CBLC siRNA caused a modest reduction in the relative be seen whether any of the known interactions with frequency of cells in both S and G2 phases of the cell CBLC are responsible for the PARP inhibitor sensitivity cycle (Supplementary Figure 2A and B). These alterations phenotype we have observed here or whether another, as were similar to those elicited by siRNA targeting BRCA2 yet unidentified, CBLC substrate mediates these effects. (Supplementary Figure 2A and B). As far as we are aware, this is the first report to Based on these analyses, we conclude that depletion link CBLC to a DNA repair related function. Compared of CLBC sensitizes diverse human cancer cells to the to other genes implicated in HR and PARP inhibitor PARP inhibitor olaparib and causes defects in cellular sensitivity, CBLC mutations are relatively rare in cancer, responses to DNA double strand breaks including Rad51 being present in approximately 2 % of 4000 human foci formation and HR repair. Furthermore, the extent tumours whose exome sequence is described on the cBio of the observed phenotypes after CBLC depletion was database [30, 31] This of course does not exclude the comparable to depletion of BRCA2, one of the key HR possibility that impaired CBLC transcription, translation factors currently being used to direct the clinical use of or reduction in CBLC protein levels by enhanced turnover PARP inhibitors. could cause sensitivity to a PARP inhibitor. Unlike its relatives within the CBL family, CBLC expression appears Discussion to be restricted to epithelial tissues [32,33]. Transgenic expression of CBLC in a mouse mammary gland causes In the work described here, we used a reanalysis impairment of cell proliferation, suggesting that CBLC of a genome-wide genetic screen [8] together with a might have a growth-inhibitory role [32]. The role that focused siRNA screen and identified and validated a CBLC plays in DNA repair by homologous recombination, novel genetic determinant of tumour cell PARP inhibitor described here, might suggest that this ubiquitin ligase sensitivity, the ubiquitin ligase CBLC [18]. siRNA could represent a candidate for a novel tumour suppressor. targeting CBLC enhanced the extent of nuclear defects It is therefore plausible that transcriptional, translational monitored as a readout in the focused screen in cells or proteolytic suppression of CBLC in tumours might be exposed to the clinical PARP inhibitor olaparib and a more common event than gene mutations. Furthermore, impaired DNA damage-induced RAD51 foci formation. in contrast to defects in genes such as BRCA1 [34], This suggested that the cause of PARP inhibitor sensitivity constitutive Cblc deficiency is neither embryonically in cells depleted of CBLC might be defective homologous lethal in mice nor overtly harmful in somatic cells of adult recombination. This hypothesis was also supported by Cblc deficient animals [33]. our experiments using a synthetic HR DNA substrate, Given the recent regulatory approval in the USA and the results of which were consistent with the concept of Europe for olaparib (now named Lynparza) as a treatment homologous recombination being indeed defective when for BRCA1/2-deficient ovarian tumours [5], the functional CBLC is silenced. assessment and validation of candidate biomarkers is most CBLC encodes an E3 ubiquitin ligase of the CBL timely. Apart from the emerging sensitivity biomarkers family, whose other members include CBL (also known identified through synthetic lethal interactions with as c-CBL) and CBLB [18]. Like CBL and CBLB, CBLC PARP inhibitors [1,2,7,8,12,34-38], several molecular has a highly conserved N-terminus that encompasses a determinants of enhanced resistance in BRCA-deficient phosphotyrosine binding domain and a catalytic ring finger tumours, such as revertant BRCA gene mutations, loss domain [22, 23]. Compared to CBL and CBLB, relatively of p53BP1 or JMJD1C have been reported and their little is known about the function of CBLC. Similar to predictive value requires careful evaluation [2,34,39-41]. CBL and CBLB, CBLC binds SH3 (SRC Homology 3) Similarly, it remains to be seen whether CBLC modulates domain containing proto-oncogenic tyrosine kinases such the clinical response to PARP inhibitors, but the work as EGFR (Epidermal Growth Factor Receptor), LYN, described here suggests that along with other genes that CRK, SRC and RET [22, 24, 25]. EGF-triggered activation are known to control homologous recombination, CBLC www.impactjournals.com/oncotarget 10755 Oncotarget should be considered as a candidate biomarker of response where average number of MN per siRNA spot in whole and potentially assessed in biopsy material from patients screen was 15.35. Threshold for hits in Olaparib treated enrolled on existing and future PARP inhibitor clinical cells was set as twice that number (2 x 15.35 = 30.7). As trials [2]. hits were considered siRNAs with more than 31 MN in the presence of Olaparib and 15 or less MN in DMSO control. Materials and Methods HR assay

Cell lines A synthetic repair reporter was used as previously described [8]. HeLa cells harboring a single-copy genomic Human U2OS cell line was cultured in Dulbecco’s integration of the DR-GFP reporter were transfected with Modified Eagle’s medium (DMEM) supplemented with siRNA targeting CBLC, BRCA2 or a siControl. 24 hours 10% (v/v) foetal bovine serum (Invitrogen, Carlsbad, later, cells were transfected with the I-SceI expression CA, USA) and penicillin/streptomycin (Sigma-Aldrich, vector, pcBASce [21]. Forty-eight hours later, HR St. Louis, MO, USA). Human MCF7 and Hs578T cell frequency was estimated by quantifying the frequency of lines were grown in RPMI 1640 or DMEM supplemented GFP positive cells using FACS [21]. with 2mM glutamine, 10% (v/v) foetal bovine serum and penicillin/streptomycin respectively. Immunocytochemistry

Chemicals The quantification of nuclear RAD51 foci was performed as previously described [3]. Briefly, MCF7 PARP inhibitor olaparib was purchased from cells transfected with siRNA targeting CBLC, BRCA2 Selleckchem and was dissolved in DMSO (Sigma-Aldrich, or siCONTROL were plated onto cover slips (BD St. Louis, MO, USA). Biosciences, Oxford, UK). 16 hours later, cells were exposed to 10 Gy ionizing radiation. At 0, 2, 4, 16 and 2 shRNA screen analysis hours after damage, cells were fixed, permeabilized, then immunostained with primary antibody targeting RAD51 (Santa Cruz Biotech) or phospho-yH2AX (Millipore) and Median drug effect Z score data from [8] was used. detected with a Texas red conjugated secondary antibody. Genes with median drug effect Z scores of <-2 in [8] were DAPI staining was used to detect nuclei. Nuclear foci cross referenced with gene listed in the following KEGG were visualized by confocal microscopy and a minimum groups to generate the data shown in Table 1: Ubiquitins of 100 fields of view were assessed. Each experiment was and Ubiquitin-like proteins, Ubiquitin-activating enzymes repeated twice. (E1), Ubiquitin-conjugating enzymes (E2), Ubiquitin ligases (E3), Deubiquitinating enzyme (DUB). Cell-based assays Targeted ubiquitin-proteasome siRNA-based screen Cell lines were transfected with SMARTpool siRNAs (Dharmacon, GE Healthcare) targeting BRCA2 and CBLC or siCONTROL A (Santa Cruz Biotech) using A custom-designed siRNA arrays were described RNAiMax (Invitrogen) transfection reagent. Cell lines previously [13]. U2OS cells were plated on siRNA arrays were infected with GIPZ shRNA constructs packaged as and 24 h later exposed to 10 µM Olaparib or DMSO. lentivirus and after 72hrs selected with 2μg/ml puromycin. Olaparib treated arrays were prepared in duplicate. Clonogenic survival assays were performed as After 72 h of cultivation cells were fixed and stained previously described [3]. Short-term survival assays were with Hoechst (Invitrogen, Carlsbad, CA, USA). Image performed in 96-well plates. Cells were seeded in 96-well acquisition of the siRNA arrays was described previously plates and drug was added after 24 hours. Cell viability [13]. The number of micronuclei was determined by was estimated after seven days using Cell-Titre Glo automated routine using software ImageJ (http://rsb. (Promega). Surviving fractions (SFs) were calculated and info.nih.gov/ij/). In addition to automatic scoring of drug sensitivity curves plotted as previously described [3]. micronuclei induction, manual scoring based on other nuclear and mitotic defects (multinucleation, irregularity Quantitative RT-PCR of nuclear shape, anaphase bridges) was employed. Hits were defined based on number of micronuclei (MN) in Olaparib control siRNA (siCON + Olaparib) treated cells, Cells were transfected with CBLC cDNA expression pCMV6 construct (Origene) then 24 hours later either www.impactjournals.com/oncotarget 10756 Oncotarget infected with CBLC or control shRNA or transfected identifies CDK12 as a novel determinant of PARP1/2 with CBLC or control siRNA. Quantitative RT-PCR was inhibitor sensitivity. Cancer Res. 2014; 74(1):287-297. carried out using Assay-on-Demand primer/probe sets 9. Jackson SP and Durocher D. Regulation of DNA damage (Applied Biosystems). Gene expression was calculated responses by ubiquitin and SUMO. Mol Cell. 2013; relative to the expression of GAPDH, and adjusted relative 49(5):795-807. to expression in shCONTROL or siCONTROL infected 10. Lukas J, Lukas C and Bartek J. More than just a focus: The cells. chromatin response to DNA damage and its role in genome integrity maintenance. Nat Cell Biol. 2011; 13(10):1161- Acknowledgements 1169. 11. Nishi R, Wijnhoven P, le Sage C, Tjeertes J, Galanty This work was funded by the European Y, Forment JV, Clague MJ, Urbe S and Jackson SP. Community’s Seventh Framework Programme (project Systematic characterization of deubiquitylating enzymes DDResponse, FP7/2007-2013) under grant agreement No. for roles in maintaining genome integrity. Nat Cell Biol. HEALTH-F2-2010-259893, the Danish Cancer Society, 2014; 16(10):1016-1026, 1011-1018. the Kellner Family Foundation, the Lundbeck Foundation 12. Fong PC, Boss DS, Yap TA, Tutt A, Wu P, Mergui- (R93-A8990), the Danish Council for Independent Roelvink M, Mortimer P, Swaisland H, Lau A, O’Connor Research (DFF-1331-00262), and the Grant Agency of the MJ, Ashworth A, Carmichael J, Kaye SB, Schellens JH and Czech Republic (Project No.:13-17555S). de Bono JS. Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N Engl J Med. 2009; 361(2):123-134. References 13. Moudry P, Lukas C, Macurek L, Hanzlikova H, Hodny Z, Lukas J and Bartek J. Ubiquitin-activating enzyme UBA1 is 1. De Vos M, Schreiber V and Dantzer F. The diverse required for cellular response to DNA damage. Cell Cycle. roles and clinical relevance of PARPs in DNA damage 2012; 11(8):1573-1582. repair: current state of the art. Biochem Pharmacol. 2012; 14. Doil C, Mailand N, Bekker-Jensen S, Menard P, Larsen 84(2):137-146. DH, Pepperkok R, Ellenberg J, Panier S, Durocher D, 2. Lord CJ, Tutt AN and Ashworth A. Synthetic Lethality and Bartek J, Lukas J and Lukas C. RNF168 binds and amplifies Cancer Therapy: Lessons Learned from the Development of ubiquitin conjugates on damaged chromosomes to allow PARP Inhibitors. Annu Rev Med. 2015; 66:455-470. accumulation of repair proteins. Cell. 2009; 136(3):435- 3. Farmer H, McCabe N, Lord CJ, Tutt AN, Johnson DA, 446. Richardson TB, Santarosa M, Dillon KJ, Hickson I, Knights 15. Scheuermann JC, de Ayala Alonso AG, Oktaba K, Ly- C, Martin NM, Jackson SP, Smith GC and Ashworth A. Hartig N, McGinty RK, Fraterman S, Wilm M, Muir TW Targeting the DNA repair defect in BRCA mutant cells as a and Muller J. Histone H2A deubiquitinase activity of the therapeutic strategy. Nature. 2005; 434(7035):917-921. Polycomb repressive complex PR-DUB. Nature. 2010; 4. Bryant HE, Schultz N, Thomas HD, Parker KM, Flower 465(7295):243-247. D, Lopez E, Kyle S, Meuth M, Curtin NJ and Helleday 16. Ihnen M, zu Eulenburg C, Kolarova T, Qi JW, Manivong T. Specific killing of BRCA2-deficient tumours with K, Chalukya M, Dering J, Anderson L, Ginther C, Meuter inhibitors of poly(ADP-ribose) polymerase. Nature. 2005; A, Winterhoff B, Jones S, Velculescu VE, Venkatesan N, 434(7035):913-917. Rong HM, Dandekar S, et al. Therapeutic potential of the 5. PARP inhibitor approved, despite vote. Nat Biotechnol. poly(ADP-ribose) polymerase inhibitor rucaparib for the 2015; 33(2):116. treatment of sporadic human ovarian cancer. Mol Cancer 6. Murai J, Huang SY, Das BB, Renaud A, Zhang Y, Ther. 2013; 12(6):1002-1015. Doroshow JH, Ji J, Takeda S and Pommier Y. Trapping of 17. Khoronenkova SV, Dianova, II, Ternette N, Kessler PARP1 and PARP2 by Clinical PARP Inhibitors. Cancer BM, Parsons JL and Dianov GL. ATM-dependent Res. 2012; 72(21):5588-5599. downregulation of USP7/HAUSP by PPM1G activates p53 7. Pettitt SJ, Rehman FL, Bajrami I, Brough R, Wallberg response to DNA damage. Mol Cell. 2012; 45(6):801-813. F, Kozarewa I, Fenwick K, Assiotis I, Chen L, Campbell 18. Swaminathan G and Tsygankov AY. The Cbl family J, Lord CJ and Ashworth A. A genetic screen using the proteins: ring leaders in regulation of cell signaling. J Cell PiggyBac transposon in haploid cells identifies Parp1 Physiol. 2006; 209(1):21-43. as a mediator of olaparib toxicity. PLoS One. 2013; 19. McCabe N, Turner NC, Lord CJ, Kluzek K, Bialkowska 8(4):e61520. A, Swift S, Giavara S, O’Connor MJ, Tutt AN, Zdzienicka 8. Bajrami I, Frankum JR, Konde A, Miller RE, Rehman FL, MZ, Smith GC and Ashworth A. Deficiency in the repair of Brough R, Campbell J, Sims D, Rafiq R, Hooper S, Chen DNA damage by homologous recombination and sensitivity L, Kozarewa I, Assiotis I, Fenwick K, Natrajan R, Lord CJ, to poly(ADP-ribose) polymerase inhibition. Cancer Res. et al. Genome-wide profiling of genetic synthetic lethality www.impactjournals.com/oncotarget 10757 Oncotarget 2006; 66(16):8109-8115. mammary gland development in MMTV-CBLC transgenic 20. Bonner WM, Redon CE, Dickey JS, Nakamura AJ, mouse. In Vivo. 2009; 23(2):225-228. Sedelnikova OA, Solier S and Pommier Y. GammaH2AX 33. Griffiths EK, Sanchez O, Mill P, Krawczyk C, Hojilla and cancer. Nat Rev Cancer. 2008; 8(12):957-967. CV, Rubin E, Nau MM, Khokha R, Lipkowitz S, Hui CC 21. Saeki H, Siaud N, Christ N, Wiegant WW, van Buul PP, and Penninger JM. Cbl-3-deficient mice exhibit normal Han M, Zdzienicka MZ, Stark JM and Jasin M. Suppression epithelial development. Mol Cell Biol. 2003; 23(21):7708- of the DNA repair defects of BRCA2-deficient cells with 7718. heterologous protein fusions. Proc Natl Acad Sci U S A. 34. Bouwman P and Jonkers J. Mouse models for BRCA1 2006; 103(23):8768-8773. associated tumorigenesis: from fundamental insights to 22. Keane MM, Ettenberg SA, Nau MM, Banerjee P, Cuello M, preclinical utility. Cell Cycle. 2008; 7(17):2647-2653. Penninger J and Lipkowitz S. cbl-3: a new mammalian cbl 35. Choi YE, Battelli C, Watson J, Liu J, Curtis J, Morse AN, family protein. Oncogene. 1999; 18(22):3365-3375. Matulonis UA, Chowdhury D and Konstantinopoulos PA. 23. Ryan PE, Kales SC, Yadavalli R, Nau MM, Zhang H and Sublethal concentrations of 17-AAG suppress homologous Lipkowitz S. Cbl-c ubiquitin ligase activity is increased recombination DNA repair and enhance sensitivity to via the interaction of its RING finger domain with a LIM carboplatin and olaparib in HR proficient ovarian cancer domain of the paxillin homolog, Hic 5. PLoS One. 2012; cells. Oncotarget. 2014; 5(9):2678-2687. 7(11):e49428. 36. Oplustilova L, Wolanin K, Mistrik M, Korinkova G, 24. Tsui CC and Pierchala BA. CD2AP and Cbl-3/Cbl-c Simkova D, Bouchal J, Lenobel R, Bartkova J, Lau A, constitute a critical checkpoint in the regulation of ret signal O’Connor MJ, Lukas J, Bartek J. Evaluation of candidate transduction. J Neurosci. 2008; 28(35):8789-8800. biomarkers to predict cancer cell sensitivity or resistance 25. Kim M, Tezuka T, Tanaka K and Yamamoto T. Cbl-c to PARP-1 inhibitor treatment Cell Cycle, 2012: 11: 3837- suppresses v-Src-induced transformation through 3850. ubiquitin-dependent protein degradation. Oncogene. 2004; 37. Ha K, Fiskus W, Choi DS, Bhaskara S, Cerchietti L, 23(9):1645-1655. Devaraj SG, Shah B, Sharma S, Chang JC, Melnick AM, 26. Calco GN, Stephens OR, Donahue LM, Tsui CC and Hiebert S and Bhalla KN. Histone deacetylase inhibitor Pierchala BA. CD2-associated protein (CD2AP) enhances treatment induces ‘BRCAness’ and synergistic lethality casitas B lineage lymphoma-3/c (Cbl-3/c)-mediated Ret with PARP inhibitor and cisplatin against human triple isoform-specific ubiquitination and degradation via its negative breast cancer cells. Oncotarget. 2014; 5(14):5637- amino-terminal Src homology 3 domains. J Biol Chem. 5650. 2014; 289(11):7307-7319. 38. Majuelos-Melguizo J, Rodriguez MI, Lopez-Jimenez L, 27. Kales SC, Nau MM, Merchant AS and Lipkowitz S. Enigma Rodriguez-Vargas JM, Martin-Consuegra JM, Serrano- prevents Cbl-c-mediated ubiquitination and degradation of Saenz S, Gavard J, de Almodovar JM and Oliver FJ. PARP RETMEN2A. PLoS One. 2014; 9(1):e87116. targeting counteracts gliomagenesis through induction of mitotic catastrophe and aggravation of deficiency in 28. Desai LP, Zhou Y, Estrada AV, Ding Q, Cheng G, Collawn homologous recombination in PTEN-mutant glioma. JF and Thannickal VJ. Negative regulation of NADPH Oncotarget. 2014. oxidase 4 by hydrogen peroxide-inducible clone 5 (Hic-5) protein. J Biol Chem. 2014; 289(26):18270-18278. 39. Watanabe S, Watanabe K. Akimov V, Bartkova J, Blagoyev B, Lukas J, Bartek J. JMJD1C demethylates MDC1 to 29. Hubackova S, Krejcikova K, Bartek J, Hodny Z. IL1- regulate the RNF8 and BRCA1 mediated chromatin and TGFbeta-Nox4 signaling, oxidative stress and DNA response to DNA breaks. Nature Struct. Mol. Biol., 2013; damage response are shared features of replicative, 20:1425-33. oncogene-induced, and drug-induced paracrine ‘Bystander senescence’. Aging (Albany NY) 4:932-51, 2012. 40. Bouwman P, Aly A, Escandell JM, Pieterse M, Bartkova J, van der GH, Hiddingh, S., Thanasoula, M., Kulkarni, 30. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, A., Yang, Q., Haffty, B,G., Tommiska, J., Blomqvist, Sumer SO, Sun Y, Jacobsen A, Sinha R, Larsson E, Cerami C., Drapkin, R., Adams, D.J., Nevanlinna, H.,Bartek, J., E, Sander C and Schultz N. Integrative analysis of complex Tarsounas, M., Ganesan, S., Jonkers, J.. 53BP1 loss rescues cancer genomics and clinical profiles using the cBioPortal. BRCA1 deficiency and is associated with triple-negative Sci Signal. 2013; 6(269):pl1. and BRCA-mutated breast cancers. Nat Struct Mol Biol. 31. Naramura M, Nadeau S, Mohapatra B, Ahmad G, 2010; 17:688-95, 2010. Mukhopadhyay C, Sattler M, Raja SM, Natarajan A, Band 41. Bouwman P, Jonkers J. Molecular pathways: how can V and Band H. Mutant Cbl proteins as oncogenic drivers in BRCA-mutated tumors become resistant to PARP myeloproliferative disorders. Oncotarget. 2011; 2(3):245- inhibitors? Clin Cancer Res. 2014; 20(3):540-7. 250. 32. Fiore F, Estebe B, Gibier P, Orsoni JC, Courbard JR, Chodosh LA, Birnbaum D and Delapeyriere O. Abnormal www.impactjournals.com/oncotarget 10758 Oncotarget